The content of this blog is my personal opinion only. Although I am an employee - currently of Imagination Technologies's MIPS group, in the past of other companies such as Intellectual Ventures, Intel, AMD, Motorola, and Gould - I reveal this only so that the reader may account for any possible bias I may have towards my employer's products. The statements I make here in no way represent my employer's position, nor am I authorized to speak on behalf of my employer. In fact, this posting may not even represent my personal opinion, since occasionally I play devil's advocate.

See http://docs.google.com/View?id=dcxddbtr_23cg5thdfj for photo credits.

Wednesday, August 16, 2017

Does Vision Research Drive Deep Learning Startups? | Chris Rowen | Pulse | LinkedIn

I think that I would like this conference format.  In preference to a multitrack conference where there are multiple papers I want to watch in the same timeslot.

Half of the reason to attend a conference in person is to observe the audience questions.

I really like the idea of poster sessions for all presented papers.

Does Vision Research Drive Deep Learning Startups? | Chris Rowen | Pulse | LinkedIn:

So many teams submit worthwhile papers [to CVPR]  that it has adopted a format to expose as many people to as many papers as possible. Roughly 2500 papers were submitted this year, of which only about 30% are accepted. Even so how can people absorb more than 750 papers? All the accepted papers get into poster sessions, and those sessions are unlike any poster session I’ve seen before. You often find two or three authors surrounded by a crowd of 25 or more, each explaining the gist of the talk over and over again for all comers. Some notable papers are given a chance to be shared a rapid-fire FOUR minute summary in one of many parallel tracks. And even smaller handful of papers gets much bigger exposure – a whole TWELVE minute slot, including Q&A!   
Remarkably the format does work – the short talks serve as useful teasers to draw people to the posters. A session of short talks gives a useful cross section of the key application problems and algorithmic methods across the whole field of computer vision. That broad sampling of the papers confirms the near-total domination of computer vision by deep learning methods. 

'via Blog this'

Sequences sunburst

Sequences sunburst:

'via Blog this'

I love data visualization. I love AI.

I want to love this visualization (of papers at the CVPR conference), but my wanna love is outweighed by my "I hate data viz that is too low bandwidth".

You can only "understand" this visualization interactively, by flying over the pie chart slices to see the labels. For somebody like me, who tends to absorb a whole picture all at once, this is much slower than presenting the piechart with labels or a legend.  Even if the legend is just color coded, because the labels don't fit.  But if the labels can be spatially attached, by proximity or by arrows, all the better.

I don't really have a photographic memory, but I have a cartoon-ographic spatial memory.  I remember pictures, usually highlighting the important areas. Sometimes my visual memory is augmented by time domain "zooming" into areas of interest.

This flyover graphic requires time domain sequential memory just for the first level absorption.  At the very least it is slower; but I suspect it also displaces the use of time domain memory for deeper understanding.

I am impelled to blog this because it is so bloody easy to make this visualization higher bandwidth.

It's a nested pie chart.  There is room for the labels for most slices of the innermost pie ring.  Even on the outermost pie ring.   I.e. without even changing the graphic most of the segments could be labelled, with interaction to dive into the segments too small for such trivial labelling.

And if you have the ability to explode sections - dynamically redraw - even more so.

I have probably made an enemy here, if this ever gets back to the author. (If you are the author, I would love talk to you.  If only to thank you for the raw data wrapped in this visualization.)

I have probably also dated myself, because video presentations are more and more the fashion.

Yes: I am the sort of person who hates watching videos, because I can read papers or slidesets faster than videos.   When I watch videos, I like fastplay and fast forward. I especially like video players that  recognize slide boundaries, so that I  can jump from slide to slide - and then only backtrack when it seems likely to have interesting discussion.

  • We can talk faster than we can type
  • But we can read faster than we can listen. Or watch video.

So, which is more important?

Sunday, August 13, 2017

Commutative FP Addition and Multiplication

In computer arithmetic, floating point addition and multiplication cannot be associative, because of rounding.

I.e. (A+B)+C != A+(B+C)

But... FP add and multiply are frequently also not commutative, in floating point instruction sets.  Not fundamentally, not because of rounding, but because of NaN and other special value propagation.


   FADD fd := fa + fb      !=       FADD fd := fb + fa    

because, if both operands are QNaNs, the instruction set may be defined to propagate the QNaN in the first operand.

This simple rule allows compiler control - but it breaks commutativity.

Losing commutativity has several downsides:

a) the instruction set cannot use operand order to provide extra information, essentially an extra opcode bit.

b) it means that a compiler cannot freely reverse operands.

Does not matter for

   FADD fd := fa + fb      !=       FADD fd := fb + fa    

But does matter for srcdst:

   FADD fa +=fb      !=       FADD fb += fa    


Other NaN propagation rules would support commutativity.

E.g. choose the NaN whose value is smallest, if the entire NaN is interpreted as an integer bitpattern.  Or largest. Signed or unsigned.

(This may be appropriate if something like a line number is encoded in the NaN.)

E.g. merge NaN operands in some way.

Tuesday, August 08, 2017

EverNote encrypts more flexibly than OneNote

Thinking about storing something semi-private in my OneNote  - nothing really important, just some medical appointments.

I've used both EverNote and OneBNote, but am currently using OneNote.

Comparisons: in this regard - encryption - EverNote seems to win. (Assuming the crypto is done properly)

Lifehacker Faceoff: OneNote vs. Evernote: "In OneNote, you can encrypt entire notebook sections, but that's only for the premium (paid) Microsoft Office versions. In Evernote, select text and right-click to encrypt it."
'via Blog this'

Saturday, August 05, 2017

Google Calendar - won't show old events

Google is missing an opportunity to be useful to people, end-users.

A calendar of future events naturally segues into a view of a journal of old events.

But I can't seem to persuade GCal to show me some events from 2011.

(Maybe it's there, and I can't find it: same diff)

Monday, July 17, 2017

ISO a decent PIM - I wish I had InfoCentral again!!

As I keep looking for a better PIM, a better to do list manager, a better cal...: " I keep looking for a better PIM, a better to do list manager, a better calendar"

'via Blog this'

closest I have ever come to a PIM that made me happy was InfoCentral on my Compaq Concerto running
Windows for Pen Computing. Circa 1994. Then later at UW Madison 1996-2000, on
another pen/tablet - I think a Toshiba or Epson?  I have had so many…

I have
blogged about this before: https://plus.google.com/+AndyGlew/posts/jDTJxgJ78F3

a description of InfoCentral: http://www.macros.koenecke.us/InfoCentral/whyic.html

The first and foremost advantage of InfoCentral, of course, is its
linking technology. Using this, any object
be it a person, organization, event, task, file on disk, or a
custom-created object
can be
linked to any other object. Further, the links themselves are objects.

From <http://www.macros.koenecke.us/InfoCentral/whyic.html>

I loved
InfoCentral's flexible but semi-structured format.
  • There were obnnjects and
    connections between objects.
  • Connections were themselves
    objects, and could have properties
    • Connections could appear
      assymmetrically  objects to which
      they were attached: eg father/daughter
    • Connections could have date
      ranges.  E.g. you could record an
      old address, but mark it no longer valid for searching
  • Objects had types, fields
    • I liked this - sometimes
    • Although I often found it
      annoying, eg when non-European name formats did not match up to
      InfoCentral's templates
    • IIRC you could hide any
      empty fields - not sure aboyt that, but obvious
  • You could create subtrees at
    any point - I thought of it as "shaking" the tree.

I think
of InfoCentral as a network database.

But it
was fairly obvious that InfoCentral was implemented, or at least could be
implemented, as a set of relational tables cross linked to each other.
     The main reason I haven't reimplemented
this is that at the time RDBMSes were not that common.  But nowadays, with SQLite, much easier.

My main
problem with InfoCentral was that it could not handle pen or bitmaps. 
     But I worked around this by linking
InfoCentral to … I can't remember the name, Inkwriter? AHV?
software that I believe Microsoft acquired, and which I think may have evolved
into OneNote.

Mainly, I
used InfoCentral nodes as a superstructure - more than a table of contents, but
less that fully integrated - to the "digital paper" or
the-software-whose-name-I-cannot remember.

could not create a TODO list by hand in the open notes, and then have
InfoCentral know ab out it.

better than nothing.

The best
PIM setup I have had to date.

Why do I
no longer use InfoCentral/Ink?   I
stopped using pen computers for a few years - hard for pair programming. I
returned to pen computers in 2002 at AMD: one day In started writing the K10
Spec, got frustrated, and at lunchtime I went out and bought a tablwet PC, a
Toshiba P4400!

But I
think that InfoCentral and this Inkwriter?? Were no longer available.

Microsoft bought the Inkwriter?? Company, and closed it down.  MS may have made either InfoCentral or
inkwriter?? Availble freely, but binary, not source.

Why don't
I revive InfoCentral?
  • I should
    • I think that I am egging
      myself on to do this uin this note.
  • But I need the digital paper
InfoCentral is fairly straightforward

Why don't
I revive Inkwriter??
  • Much more UI stuff.
  • If I could get OneNote
    convenient for starting single files or linking back…
  • Years ago, UNIXes had little
    pen/touch support.

It would
be fun to write myself - but it will take a long time to get to a usable placed

I really
do just wish it was out there in a form that I could buy.

Saturday, July 08, 2017

Waterproofed Fitbit Blaze

Waterproofed Fitbit Blaze: "Waterproofed Fitbit Blaze"

'via Blog this'

I just googled "waterproof Fitbit watch", since I have been waiting for the Fitbit Blaze replacement that is supposedly swimmable. Search turned up WaterFi, a company that does aftermarket waterproofing. 

WaterFi has a good rep, both on web and by personal experience - I have a waterfi'ed iPod Shuffle.

WaterFi sells a number of aftermarket waterproofed trackers, ranging from Fitbit Charge and Charge 2 HR to Fitbit Blaze.

They don't change the software, so no "swim mode". But reports say that they count "strokes" as "steps", which is good enough. 

The waterproofing closes up the air pressure port, so the altimeter "flights climbed" feature breaks. Minor sadness: living on a hillside, vertical is a useful exercise metric, correlates to intensity; but Fitbit does not do much with vertical, doesn't use for challenges, etc.

329$ for the large waterproofed Blaze, 300$ small, vs 199$ list, sometimes 150$ on sale. Since WaterFi offers a 99$ waterproofing service, it might be more cost effective to buy from FitBit or a reseller, and then send to WaterFi.

I am especially interested in the WaterFi'ed Fitbit Blaze, which has standard watchstraps (22/23/24mm?b - many for sale on Amazon) . "Leaked" photos of the (late, not yet shipping) new Fitbit watch seem to show that it does not use standard watchstraps.

---+ WaterFi'ed iPod Shuffle

A few years ago I bought a WaterFi-ed swimmable iPod Shuffle, which I used to listen to podcasts while swimming for quite a while. Notes: music fine; podcasts okay while doing breaststroke, but hard to hear doing crawl, and always miss something during tumble turns.) Still have, still works; I only stopped using (a) swimmer's knee (breaststroke), (b) got Fitbit, stopped swimming (c) it's a pain to sync podcasts to - modern podcatchers don't seem to handle non-phone devices like iPods.  But I still use occasionally, when I have a lot of podcasts or audiobooks to upload, amortizing the hassle of doing so.

Prompts thoughts: one of the things that attracts me about an Apple Watch is that it can supposedly play podcasts from the watch even when not carrying phone.  But I'll bet this doesn't work in the water (?)